Capability · OCR alpha

Structured data does not always start structured.

Extract text, line locations and model scores from PNG and JPEG images. Use the recognized text in your own review, mapping and validation workflow.

Try OCR →Read the guide

An input capability for your workflow

Run recognition locally, inspect the result and send its text to the Importer or Converter SDK playground. OCR does not automatically extract a business schema or reconstruct tables.

Know the current scope

PNG and JPEG input. PDF ingestion is not part of this experience. Accuracy and reading order depend on your image and selected model.

Package availability

Preview · npm release pending

@diakrio/ocr

This scoped package has no verified public npm release as of 2026-09-18. You can try the locally built runtime on this site while the release is prepared.

Free for personal and commercial application use. License and notices.

Loading page…